7 UI, Workflows & Features
9 * "branch --delete" has "branch -d" but "push --delete" does not.
11 * "git blame" learned to produce the progress eye-candy when it takes
12 too much time before emitting the first line of the result.
14 * "git grep" can now be configured (or told from the command line)
15 how many threads to use when searching in the working tree files.
17 * Some "git notes" operations, e.g. "git log --notes=<note>", should
18 be able to read notes from any tree-ish that is shaped like a notes
19 tree, but the notes infrastructure required that the argument must
20 be a ref under refs/notes/. Loosen it to require a valid ref only
21 when the operation would update the notes (in which case we must
22 have a place to store the updated notes tree, iow, a ref).
24 * "git grep" by default does not fall back to its "--no-index"
25 behaviour outside a directory under Git's control (otherwise the
26 user may by mistake end up running a huge recursive search); with a
27 new configuration (set in $HOME/.gitconfig--by definition this
28 cannot be set in the config file per project), this safety can be
31 * "git pull --rebase" has been extended to allow invoking
34 * "git p4" learned to cope with the type of a file getting changed.
36 * "git format-patch" learned to notice format.outputDirectory
37 configuration variable. This allows "-o <dir>" option to be
38 omitted on the command line if you always use the same directory in
41 * "interpret-trailers" has been taught to optionally update a file in
42 place, instead of always writing the result to the standard output.
44 * Many commands that read files that are expected to contain text
45 that is generated (or can be edited) by the end user to control
46 their behaviour (e.g. "git grep -f <filename>") have been updated
47 to be more tolerant to lines that are terminated with CRLF (they
48 used to treat such a line to contain payload that ends with CR,
49 which is usually not what the users expect).
51 * "git notes merge" used to limit the source of the merged notes tree
52 to somewhere under refs/notes/ hierarchy, which was too limiting
53 when inventing a workflow to exchange notes with remote
54 repositories using remote-tracking notes trees (located in e.g.
55 refs/remote-notes/ or somesuch).
57 * "git ls-files" learned a new "--eol" option to help diagnose
60 * "ls-remote" learned an option to show which branch the remote
61 repository advertises as its primary by pointing its HEAD at.
63 * New http.proxyAuthMethod configuration variable can be used to
64 specify what authentication method to use, as a way to work around
65 proxies that do not give error response expected by libcurl when
66 CURLAUTH_ANY is used. Also, the codepath for proxy authentication
67 has been taught to use credential API to store the authentication
68 material in user's keyrings.
70 * Update the untracked cache subsystem and change its primary UI from
71 "git update-index" to "git config".
73 * There were a few "now I am doing this thing" progress messages in
74 the TCP connection code that can be triggered by setting a verbose
75 option internally in the code, but "git fetch -v" and friends never
76 passed the verbose option down to that codepath.
78 * Clean/smudge filters defined in a configuration file of lower
79 precedence can now be overridden to be a pass-through no-op by
80 setting the variable to an empty string.
82 * A new "<branch>^{/!-<pattern>}" notation can be used to name a
83 commit that is reachable from <branch> that does not match the
87 Performance, Internal Implementation, Development Support etc.
89 * Add a framework to spawn a group of processes in parallel, and use
90 it to run "git fetch --recurse-submodules" in parallel.
92 * A slight update to the Makefile to mark "phoney" targets
95 * In-core storage of the reverse index for .pack files (which lets
96 you go from a pack offset to an object name) has been streamlined.
98 * d95138e6 (setup: set env $GIT_WORK_TREE when work tree is set, like
99 $GIT_DIR, 2015-06-26) attempted to work around a glitch in alias
100 handling by overwriting GIT_WORK_TREE environment variable to
101 affect subprocesses when set_git_work_tree() gets called, which
102 resulted in a rather unpleasant regression to "clone" and "init".
103 Try to address the same issue by always restoring the environment
104 and respawning the real underlying command when handling alias.
106 * The low-level code that is used to create symbolic references has
107 been updated to share more code with the code that deals with
110 * strbuf_getline() and friends have been redefined to make it easier
111 to identify which callsite of (new) strbuf_getline_lf() should
112 allow and silently ignore carriage-return at the end of the line to
113 help users on DOSsy systems.
115 * "git shortlog" used to accumulate various pieces of information
116 regardless of what was asked to be shown in the final output. It
117 has been optimized by noticing what need not to be collected
118 (e.g. there is no need to collect the log messages when showing
119 only the number of changes).
121 * "git checkout $branch" (and other operations that share the same
122 underlying machinery) has been optimized.
124 * Automated tests in Travis CI environment has been optimized by
125 persisting runtime statistics of previous "prove" run, executing
126 tests that take longer before other ones; this reduces the total
130 Also contains various documentation updates and code clean-ups.
136 Unless otherwise noted, all the fixes since v2.7 in the maintenance
137 track are contained in this release (see the maintenance releases'
140 * An earlier change in 2.5.x-era broke users' hooks and aliases by
141 exporting GIT_WORK_TREE to point at the root of the working tree,
142 interfering when they tried to use a different working tree without
143 setting GIT_WORK_TREE environment themselves.
144 (merge df1e6ea nd/stop-setenv-work-tree later to maint).
146 * The "exclude_list" structure has the usual "alloc, nr" pair of
147 fields to be used by ALLOC_GROW(), but clear_exclude_list() forgot
148 to reset 'alloc' to 0 when it cleared 'nr' to discard the managed
150 (merge 2653a8c nd/dir-exclude-cleanup later to maint).
152 * Paths that have been told the index about with "add -N" are not
153 quite yet in the index, but a few commands behaved as if they
154 already are in a harmful way.
155 (merge 4d55200 nd/ita-cleanup later to maint).
157 * "git send-email" was confused by escaped quotes stored in the alias
158 files saved by "mutt", which has been corrected.
159 (merge 2c510f2 ew/send-email-mutt-alias-fix later to maint).
161 * A few unportable C construct have been spotted by clang compiler
163 (merge a0df2e5 jk/clang-pedantic later to maint).
165 * The documentation has been updated to hint the connection between
166 the '--signoff' option and DCO.
167 (merge b2c150d dw/signoff-doc later to maint).
169 * "git reflog" incorrectly assumed that all objects that used to be
170 at the tip of a ref must be commits, which caused it to segfault.
171 (merge aecad37 dk/reflog-walk-with-non-commit later to maint).
173 * The ignore mechanism saw a few regressions around untracked file
174 listing and sparse checkout selection areas in 2.7.0; the change
175 that is responsible for the regression has been reverted.
176 (merge 8c72236 nd/exclusion-regression-fix later to maint).
178 * Some codepaths used fopen(3) when opening a fixed path in $GIT_DIR
179 (e.g. COMMIT_EDITMSG) that is meant to be left after the command is
180 done. This however did not work well if the repository is set to
181 be shared with core.sharedRepository and the umask of the previous
182 user is tighter. They have been made to work better by calling
183 unlink(2) and retrying after fopen(3) fails with EPERM.
184 (merge ea56518 js/fopen-harder later to maint).
186 * Asking gitweb for a nonexistent commit left a warning in the server
189 Somebody may want to follow this up with an additional test, perhaps?
190 IIRC, we do test that no Perl warnings are given to the server log,
191 so this should have been caught if our test coverage were good.
192 (merge a9eb90a ho/gitweb-squelch-undef-warning later to maint).
194 * "git rebase", unlike all other callers of "gc --auto", did not
195 ignore the exit code from "gc --auto".
196 (merge 8c24f5b jk/ok-to-fail-gc-auto-in-rebase later to maint).
198 * Many codepaths that run "gc --auto" before exiting kept packfiles
199 mapped and left the file descriptors to them open, which was not
200 friendly to systems that cannot remove files that are open. They
201 now close the packs before doing so.
202 (merge d562102 js/close-packs-before-gc later to maint).
204 * A recent optimization to filter-branch in v2.7.0 introduced a
205 regression when --prune-empty filter is used, which has been
207 (merge 1dc413e jk/filter-branch-no-index later to maint).
209 * The description for SANITY prerequisite the test suite uses has
210 been clarified both in the comment and in the implementation.
211 (merge 719c3da jk/sanity later to maint).
213 * "git tag" started listing a tag "foo" as "tags/foo" when a branch
214 named "foo" exists in the same repository; remove this unnecessary
215 disambiguation, which is a regression introduced in v2.7.0.
216 (merge 0571979 jk/list-tag-2.7-regression later to maint).
218 * The way "git svn" uses auth parameter was broken by Subversion
220 (merge 0b66415 ew/svn-1.9.0-auth later to maint).
222 * The "split" subcommand of "git subtree" (in contrib/) incorrectly
223 skipped merges when it shouldn't, which was corrected.
224 (merge 933cfeb dw/subtree-split-do-not-drop-merge later to maint).
226 * A few options of "git diff" did not work well when the command was
227 run from a subdirectory.
228 (merge a97262c nd/diff-with-path-params later to maint).
230 * The command line completion learned a handful of additional options
231 and command specific syntax.
232 (merge fa4b5e3 jk/completion-rebase later to maint).
233 (merge f7c2e1a pw/completion-show-branch later to maint).
234 (merge d7d4ca8 pw/completion-stash later to maint).
235 (merge e6414b4 tb/complete-word-diff-regex later to maint).
237 * dirname() emulation has been added, as Msys2 lacks it.
238 (merge e7d5ce8 js/dirname-basename later to maint).
240 * The underlying machinery used by "ls-files -o" and other commands
241 have been taught not to create empty submodule ref cache for a
242 directory that is not a submodule. This removes a ton of wasted
244 (merge a2d5156 jk/ref-cache-non-repository-optim later to maint).
246 * "git worktree" had a broken code that attempted to auto-fix
247 possible inconsistency that results from end-users moving a
248 worktree to different places without telling Git (the original
249 repository needs to maintain backpointers to its worktrees, but
250 "mv" run by end-users who are not familiar with that fact will
251 obviously not adjust them), which actually made things worse
253 (merge 618244e nd/do-not-move-worktree-manually later to maint).
255 * The low-level merge machinery has been taught to use CRLF line
256 termination when inserting conflict markers to merged contents that
257 are themselves CRLF line-terminated.
258 (merge 15980de js/xmerge-marker-eol later to maint).
260 * "git push --force-with-lease" has been taught to report if the push
261 needed to force (or fast-forwarded).
262 (merge b2e93f8 aw/push-force-with-lease-reporting later to maint).
264 * The emulated "yes" command used in our test scripts has been
265 tweaked not to spend too much time generating unnecessary output
266 that is not used, to help those who test on Windows where it would
267 not stop until it fills the pipe buffer due to lack of SIGPIPE.
268 (merge 6129c93 js/test-lib-windows-emulated-yes later to maint).
270 * Other minor clean-ups and documentation updates
271 (merge 99487cf ss/user-manual later to maint).
272 (merge e914ef0 ew/for-each-ref-doc later to maint).
273 (merge 36fc7d8 sg/t6050-failing-editor-test-fix later to maint).
274 (merge 60253a6 ss/clone-depth-single-doc later to maint).
275 (merge bd02e97 lv/add-doc-working-tree later to maint).